A fuzzy classifier system that generates fuzzy if-then rules for pattern classification problems

Abstract

We propose a fuzzy classifier system that can automatically generate fuzzy if-then rules from numerical data (i.e., from training patterns) for multi-dimensional pattern classification problems. Classifiers in our approach are fuzzy if-then rules such as "If x/sub p1/ is small and x/sub p2/ is large then classify x/sub p/ as Class 2". The proposed classifier system can find a compact rule set by attaching large fitness values to such fuzzy if-then rules that can correctly classify many training patterns. That is, only fuzzy if-then rules with large fitness values are selected to construct a compact fuzzy system with high classification performance.